Social capital development on interest-based networks: examining its antecedents, process, and consequences

The study addresses whether interest-based social networking sites (SNSs) generate social capital and through which activities, how bridging (weak ties) and bonding (strong ties) social capital develop on such platforms, and whether social capital enhances site vitality as reflected by users’ sense of belonging. Unlike relationship-based networks where users focus on maintaining or building interpersonal ties, interest-based networks center on content exchange around shared topics. The authors propose that on interest-based SNSs, active online interactions (OIs) with content and with humans may differentially foster bridging and bonding social capital, and that these forms of social capital contribute to a sense of belonging. The research articulates three questions: (1) whether the sites breed social capital and via what online activities; (2) how strong and weak ties develop from those activities; and (3) whether social capital contributes to site vitality. Corresponding hypotheses test the roles of OIs with content and humans in predicting bridging; a sequential path from bridging to bonding; and associations between OIs, social capital, and sense of belonging.

The paper grounds itself in social capital theory, defining social capital as resources embedded in social networks accessed through ties (Lin, 2001). It distinguishes structural, relational, and cognitive dimensions and emphasizes the relational dimension (bridging and bonding) as both cause and outcome across domains. Bridging social capital involves loose, inclusive ties facilitating information diffusion, whereas bonding involves strong, exclusive ties providing substantial support (Granovetter, 1982; Putnam, 2000). Prior work largely focuses on relationship-based networks (Facebook, LinkedIn, WeChat), showing active behaviors (broadcasting, chatting) build social capital while passive browsing is less effective. The review highlights two distinctive elements of SNS activity: OIs with content (creation, sharing, favoriting) and OIs with humans (direct communications). It posits that content-focused activity can expose users to wider audiences and attract like-minded connections, potentially fostering bridging, while human-directed interactions directly build ties. The authors theorize a sequential relationship where bridging precedes and facilitates bonding, supported by management studies noting positive associations between these forms. Finally, literature connects social capital to a sense of belonging, a critical indicator of community vitality on interest-based networks; both OIs and accumulated social capital are expected to enhance belonging.

Design and platform: A cross-sectional online survey was conducted on Douban.com, a large Chinese interest-based network focused on cultural content (movies, books, music). The platform supports both OIs with humans (e.g., commenting on reviews, forum discussions, private chats) and OIs with content (e.g., rating, marking, reviewing).

Participants and procedure: A pilot with 73 Douban users established reliability and construct validity (all Cronbach’s alpha > 0.7). The main survey was fielded in June 2021 via Wenjuanxing. Recruitment announcements were posted on major Chinese social media. Participants received 3 RMB for completion. N = 624 Douban users completed the survey (58.7% female; 41.3% male). Age distribution: 2.9% below 18, 70.7% 18–25, 21.8% 26–35, 4.0% 36–45, 0.6% above 45. Education: 3.2% high school and below, 6.7% junior college, 76.8% undergraduate, 13.3% graduate. Monthly disposable income (RMB): 23.6% <1500, 45.0% 1500–2999, 16.8% 3000–4999, 8.8% 5000–7999, 5.8% ≥8000.

Measures:

• OIs with content: Three frequency items over past six months (rating, marking, reviewing); α = 0.840.
• OIs with humans: Four frequency items (participating in forum discussions, private chatting, liking, commenting on others’ posts); α = 0.785.
• Bridging social capital: Five items adapted from Williams (2006) and Ellison et al. (2007), e.g., “Interactions on Douban.com make me interested in what people unlike me are thinking”; α = 0.876.
• Bonding social capital: Five items adapted from the same sources, e.g., “There are several people on Douban.com I trust to solve my problems”; α = 0.855.
• Sense of belonging: Three items (McMillan & Chavis, 1986; Zhao et al., 2012): “I have a sense of belonging to Douban.com.” “I feel close to other members on Douban.com.” “I am proud of being a Douban.com user.”; α = 0.900. All scales used 5-point Likert responses unless otherwise noted.

Descriptive statistics and correlations: Means (SDs): OIs with content 3.071 (0.965); OIs with humans 2.907 (0.879); Bridging 3.735 (0.771); Bonding 2.215 (0.895); Sense of belonging 3.141 (1.064). All pairwise correlations were significant at p < .01; all < 0.70, indicating no multicollinearity.

Analytic strategy: Two serial mediation models (PROCESS v3.5 Model 6) tested pathways from (a) OIs with content and (b) OIs with humans (independent variables) to sense of belonging (dependent variable) via bridging (stage-one mediator) and bonding (stage-two mediator). The alternative OI type and demographics (age, sex, education, income) were covariates. Indirect effects were assessed via 5,000-sample bias-corrected bootstrapping (95% CIs). Model fit statistics (R2 and F) were reported for each stage outcome.

• Correlations: OIs with content correlated with bridging r = 0.356 and bonding r = 0.312; OIs with humans correlated with bridging r = 0.500 and bonding r = 0.496; sense of belonging correlated with bridging r = 0.597 and bonding r = 0.550 (all p < .01).

Model with OIs with content as IV (covarying OIs with humans and demographics):

• OIs with content → Bridging: B = 0.113, SE = 0.032, p < 0.001 (significant).
• OIs with content → Bonding: B = 0.066, SE = 0.037, p = 0.078 (ns).
• Bridging → Bonding: B = 0.154, SE = 0.047, p < 0.01 (significant).
• Sense of belonging ← Bridging: B = 0.550, SE = 0.046, p < 0.001; ← Bonding: B = 0.392, SE = 0.040, p < 0.001; ← OIs with content: B = 0.051, SE = 0.037, p = 0.164 (ns).
• R2/F: Bridging R2 = 0.269, F = 37.910***; Bonding R2 = 0.271, F = 32.671***; Sense of belonging R2 = 0.506, F = 78.632***.
• Indirect effects (bootstrapped, 95% CI): • Content → Bridging → Belonging: B = 0.062, CI [0.025, 0.102] (significant). • Content → Bonding → Belonging: B = 0.026, CI [−0.003, 0.056] (ns). • Content → Bridging → Bonding → Belonging: B = 0.007, CI [0.002, 0.013] (significant).

Model with OIs with humans as IV (covarying OIs with content and demographics):

• OIs with humans → Bridging: B = 0.374, SE = 0.035, p < 0.001.
• OIs with humans → Bonding: B = 0.392, SE = 0.044, p < 0.001.
• Bridging → Bonding: B = 0.154, SE = 0.047, p < 0.01.
• Sense of belonging ← OIs with humans: B = 0.149, SE = 0.046, p = 0.001; ← Bridging: B = 0.550, SE = 0.046, p < 0.001; ← Bonding: B = 0.392, SE = 0.039, p < 0.001.
• Indirect effects (bootstrapped, 95% CI): • Humans → Bridging → Belonging: B = 0.206, CI [0.151, 0.264] (significant). • Humans → Bonding → Belonging: B = 0.154, CI [0.112, 0.201] (significant). • Humans → Bridging → Bonding → Belonging: B = 0.023, CI [0.009, 0.039] (significant).

Hypotheses summary:

• H1a supported: OIs with content predict bridging.
• H1b supported: OIs with humans predict bridging.
• H2 supported: Bridging predicts bonding (sequential development).
• H3a rejected: OIs with content do not directly predict sense of belonging.
• H3b supported: OIs with humans directly predict sense of belonging.
• H3c supported: Bridging predicts sense of belonging.
• H3d supported: Bonding predicts sense of belonging.
• RQ1a: No direct association between OIs with content and bonding (ns).
• RQ1b: OIs with humans directly predict bonding (significant).

The findings demonstrate that interest-based SNSs can cultivate social capital despite users’ primary motivation being information exchange. Active OIs with humans directly foster both bridging and bonding, while OIs with content foster bridging but not bonding directly. This supports a sequential process: users first develop weak ties (bridging) through exposure around shared interests, some of which deepen into strong ties (bonding) as interactions intensify, aligning with social penetration perspectives. Both bridging and bonding, in turn, strongly contribute to a sense of belonging—an indicator of the cognitive dimension of social capital and site vitality. Notably, OIs with humans have both direct and indirect effects on belonging, whereas content interactions impact belonging only indirectly through social capital. The results extend social capital theory by applying it to interest-based networks, identifying distinct antecedents (content- and human-focused interactions), and elucidating the interrelationship between bridging and bonding as sequential rather than purely parallel constructs. Practically, platforms should encourage both content contribution and social interactions, but also implement features that convert content engagement into relationship formation (e.g., topic-based friend recommendations), as belonging appears to be mediated by relational development. The study also suggests social and political implications, as interest-based platforms can enhance inclusion and community engagement by facilitating both bridging and bonding ties.

The study shows that on interest-based networks, social capital develops through active engagement: OIs with content and humans build bridging ties, which can subsequently nurture bonding ties. Both forms of social capital significantly enhance users’ sense of belonging, while human-directed interactions also exert a direct positive effect. This work contributes by demonstrating social capital formation in interest-based contexts, distinguishing activity-specific antecedents, and clarifying a sequential bridging-to-bonding pathway. Future work should employ longitudinal designs to confirm causal ordering, incorporate objective behavioral and network data, replicate across varied interest-based platforms (e.g., travel, education), and unpack the roles of specific activities (e.g., commenting, sharing) within the broader OI categories.

• Cross-sectional design precludes causal inference; longitudinal research is needed, especially to verify the proposed bridging-to-bonding sequence.
• Reliance on self-reported measures may introduce subjectivity and inaccuracy; future studies should integrate objective behavioral footprints and network structure data.
• Single-platform focus (Douban.com centered on cultural works) limits generalizability; replication across other interest-based contexts is recommended.
• Broad categorization of activities into OIs with content and humans; examining specific behaviors (e.g., commenting, sharing) could provide more granularity.